Stability problem

[shattered]

Comp specs are in sig.  I have had this computer for maybe 3 years with no issues.  Recently it has begun to have fatal errors on a regular basis that lead to the BSOD.  I do not know much about the event view, but here is what it logs each time this happens.

 

 

- System

 

    - Provider

      [ Name] Microsoft-Windows-Kernel-Power       [ Guid] {331C3B3A-2005-44C2-AC5E-77220C37D6B4}

      EventID 41       Version 2       Level 1       Task 63       Opcode 0       Keywords 0x8000000000000002     - TimeCreated

      [ SystemTime] 2014-11-14T22:40:05.041206200Z

      EventRecordID 254433       Correlation     - Execution

      [ ProcessID] 4       [ ThreadID] 8

      Channel System       Computer Death     - Security

      [ UserID] S-1-5-18

 

- EventData

 

    BugcheckCode 80     BugcheckParameter1 0xfffffaa0137ccff8     BugcheckParameter2 0x0     BugcheckParameter3 0xfffff88006168ebe     BugcheckParameter4 0x5     SleepInProgress false     PowerButtonTimestamp 0

 

Prior to this, the computer was stable at Prime95 for 24 hours with room temperature of 78 F.  Now, core #1 (second core) does not pass around 13 minutes of the CPU stress test.  The other three cores passed up to 14 hours.  Memtest86+ reveals no issues.  The crashes happen at various times such as browsing the web or playing a game such as Dishonored.  All four cores run at 82 F at idle.  At full load they got to 178 F. 

 

My question is how did the CPU go from being stable to unstable?  Is there anything that can be done to "fix" the CPU or could it be some other issue that is affecting that one core?  Any ideas would be greatly appreciated.  Sorry if I left out crucial information.  Thanks.

[Tidosho]

My question is how did the CPU go from being stable to unstable?  Is there anything that can be done to "fix" the CPU or could it be some other issue that is affecting that one core?  Any ideas would be greatly appreciated.  Sorry if I left out crucial information.  Thanks.

 

Integrated electronics like IC's and CPU's can either fail suddenly, or gradually. My guess is if it's just one core causing it, your CPU may be faulty. A few questions:

 

1. Is your heatsink firmly/correctly attached, fan running at the correct speed, any fan speed adjustment under load working correctly? Some CPU fans speed up depending on load like they do laptops. Does your thermal paste get renewed regularly? I know your other cores can go 14 hours, but throwing this out there as I can't see your machine so I need to think logically.

 

2. Are you sure your PSU is good, mainly the VCORE voltage of the CPU? Use HWMonitor, and monitor your voltages. ATX specification specifies that the 3.3V, 5V, and 12V rails shall supply power within +-5% of the rated voltage, so the +12v can read as low as 11.4 to 12.6v and be within tolerances, for example. A PSU problem could be a cause, as it happens when all cores are enabled and stressed, but not when Core #1 is disabled, especially if the CPU core voltage is iffy.

 

3. Have you tried disabling all cores but #1 and re-running the test on just one, like running an engine on less than full cylinders (my fave analogy!)

[shattered]

Thanks for the reply.  I am not sure what I messed up on while trying to run Memtest, but I ran it again and had over 6700 errors in a 20 min span.  It had a 0 next to the last column which states "CPU" so I am not sure if the bad memory could cause the core 1 of the CPU to fail on Prime95 or if the CPU could be bad also.

 

I checked the voltages and the lowest the VCore got was .840 but it fluctuates a lot up to 1.096 over a 5 min period.  I am not sure what a normal level is for that value but the default min in HWMonitor is .808v.  I checked the heat sink and it is still attached firmly.  The fan is working normal at 800rpm idle.  The thermal paste I used was Arctic Silver mx4, and I haven't changed it since I installed the CPU 3 or so years ago.

 

I will now try to isolate which stick(s) of ram are bad and then use just the good ones while running Prime95 to see if the CPU still has errors. 

 

 

[Tidosho]

I checked the voltages and the lowest the VCore got was .840 but it fluctuates a lot up to 1.096 over a 5 min period.  I am not sure what a normal level is for that value but the default min in HWMonitor is .808v.

 

That is normal. The voltage drop is called V Droop, it's a protection feature so voltage spikes don't kill the CPU, your CPU VRM's (Voltage Regulator Modules) will take care of it. Idling, you'll get a higher VCORE, while at 100% load it will be a lower voltage. 1.24v is the standard VCore voltage for your i5 2500K. I recommend your thermal paste be changed at least every 12 months, especially if stressed a lot.